David Romero-Organvidez has presented “An Instrument for Measuring Variability Thinking” at VARIABILITY 2026, held in Limassol, Cyprus. The work is co-authored with Jesús Moreno-León, José A. Galindo and David Benavides, all from the I3US Institute at the University of Seville.
Variability is a central concept in software product lines. Engineers must identify options, understand dependencies and analyze configurations. However, while the community has produced a wide range of techniques, tools and analyses, there was no instrument to measure the ability to reason about variation. Without it, it is difficult to compare profiles, assess skills or evaluate the impact of teaching.
To address this gap, the authors propose the Test of Variability Thinking (TVT), a brief visual test of ten questions that takes around 15 minutes and requires no prior technical knowledge. Inspired by existing computational thinking tests, the TVT uses a Playmobil-inspired medieval castle domain to represent options, dependencies and constraints, so that participants reason about variability rather than about notations or tools. The items are structured following Apel et al.’s classic SPL framework and are all derived from a single reference UVL model, which ensures internal consistency.
The test was refined in three phases: an internal validation with ten researchers of varying SPL expertise, a focus group with seven software engineering students, and a classroom administration with 88 fourth-year Software Engineering students. Results show a good spread of scores, items with generally good discrimination, and no significant differences associated with gender, familiarity with Playmobil or prior configuration experience. The resulting version 0.9 offers initial evidence that variability thinking can be measured in a technology-independent way.
Looking ahead, the authors plan to expand and validate the instrument with larger samples and to develop a family of tests, a TVT product line, adapted to different audiences and educational stages.
The test form, the reference UVL model and the statistical analysis are openly available:
Test form (PDF): https://doi.org/10.5281/zenodo.17910732
UVL model on UVLHub: https://www.uvlhub.io/doi/10.5281/zenodo.17910370
Statistical analysis: https://doi.org/10.5281/zenodo.19352175
This work was partially supported by FEDER, the Ministry of Science, Innovation and Universities, Junta de Andalucía, the State Research Agency and CDTI through the projects Data-pl (PID2022-138486OB-I00) and SENSOLIVE (PLSQ_00162).